In [1]:
import pandas as pd
import numpy as np
import csv
import matplotlib.pyplot as plt
from sklearn import metrics
from sklearn.metrics import classification_report, accuracy_score, precision_score, recall_score, f1_score
from sklearn.metrics import r2_score, mean_squared_error, mean_absolute_error,mean_absolute_percentage_error
from sklearn.preprocessing import MinMaxScaler
In [2]:
filename = "predicted_series_{0}_block{1}_w{2}.csv"
models = ['rnn','lstm']
blocks = ['1','10','20','30']
win = ['50','20']
In [3]:
test_data_file = pd.read_csv("test_data3.csv")
names = test_data_file.columns.tolist()
In [4]:
def mean_std(test_list):
    mean = sum(test_list) / len(test_list) 
    variance = sum([((x - mean) ** 2) for x in test_list]) / len(test_list) 
    res = variance ** 0.5
    return [mean, res]
In [5]:
from statistics import mean
In [6]:
R2score_tot = []
RMSE_tot = []
MAPE_tot = []
MAE_tot = []
MAPE_labels = []
files = []
windows = []
b = []
m = []
c=0
high_acc = []
good = []
mid = []
bad = []
for model in models:
    for block in blocks:
        for w in win:
            c+=1
            file = filename.format(model,block,w)
            testfile = "test_data{0}.csv".format(blocks.index(block)+1)
            windows.append(w)
            m.append(model)
            b.append(block)
            predicted_series_rnn = pd.read_csv(file)
            test_data_file = pd.read_csv(testfile)
            files.append(file)
            print(file, testfile)
            R2score = []
            RMSE = []
            MAPE = []
            MAE = []
            for name in names:
                test = test_data_file[name][int(w):].tolist()
                pred = predicted_series_rnn[name].tolist()
                #plot(pred,test)
                R2score.append(r2_score(test, pred))
                RMSE.append(np.sqrt(mean_squared_error(test, pred)))
                MAPE.append(mean_absolute_percentage_error(test, pred))
                MAE.append(mean_absolute_error(test, pred))
                
            high = 0
            good_ = 0
            mid_ = 0
            bad_ = 0
            for i in range(len(MAPE)):
                if MAPE[i]<=0.10:
                    high+=1
                elif MAPE[i]<=0.20 and MAPE[i]>0.10:
                    good_+=1
                elif MAPE[i]<=0.50 and MAPE[i]>0.20:
                    mid_+=1
                elif MAPE[i]>0.50:
                    bad_+=1
                    
            high_acc.append(high)
            good.append(good_)
            mid.append(mid_)
            bad.append(bad_)
            R2score_tot.append(mean_std(R2score))
            RMSE_tot.append(mean_std(RMSE))
            MAPE_tot.append(mean_std(MAPE))
            MAE_tot.append(mean_std(MAE))

predicted_series_rnn_block1_w50.csv test_data1.csv
predicted_series_rnn_block1_w20.csv test_data1.csv
predicted_series_rnn_block10_w50.csv test_data2.csv
predicted_series_rnn_block10_w20.csv test_data2.csv
predicted_series_rnn_block20_w50.csv test_data3.csv
predicted_series_rnn_block20_w20.csv test_data3.csv
predicted_series_rnn_block30_w50.csv test_data4.csv
predicted_series_rnn_block30_w20.csv test_data4.csv
predicted_series_lstm_block1_w50.csv test_data1.csv
predicted_series_lstm_block1_w20.csv test_data1.csv
predicted_series_lstm_block10_w50.csv test_data2.csv
predicted_series_lstm_block10_w20.csv test_data2.csv
predicted_series_lstm_block20_w50.csv test_data3.csv
predicted_series_lstm_block20_w20.csv test_data3.csv
predicted_series_lstm_block30_w50.csv test_data4.csv
predicted_series_lstm_block30_w20.csv test_data4.csv
In [7]:
df = pd.DataFrame({
    'Files': files,
    'Block':b,
    'Window':windows,
    'Model': m,
    'High': high_acc,
    'Good': good,
    'Mid': mid,
    'Bad': bad
})
In [8]:
df_round = df.round(2)
df_round.to_csv("mape_labels.csv")
In [9]:
R2_mean = []
R2_std = []
for i in R2score_tot:
    print(i, i[1]/i[0])
    R2_mean.append(i[0])
    R2_std.append(i[1])

[0.7905424959699804, 0.1504867927385568] 0.1903588908954381
[0.7897788660808004, 0.12580194098077263] 0.1592875504570695
[0.9702673819021994, 0.04010606969724329] 0.04133506953373589
[0.9644707545270543, 0.028277614551443286] 0.029319307421933935
[0.9819790827441726, 0.024656725571361164] 0.025109216687647908
[0.9778311480320238, 0.027103317326623456] 0.02771778888530132
[0.9824152815270502, 0.026395411708256322] 0.026867875739094497
[0.9788110570555104, 0.028352615955096937] 0.02896638299161449
[0.7872163549790705, 0.1311912057821846] 0.16665203276381693
[0.7922328482554688, 0.09987584608149362] 0.12606880199606035
[0.9761188517741332, 0.0324048214051848] 0.03319761865708034
[0.9674927054050065, 0.04610109096541769] 0.04765006568821529
[0.9820570574218511, 0.03720732624419269] 0.03788713289416336
[0.9767062163791912, 0.022117303792646207] 0.022644786550697557
[0.981517241647321, 0.038326014365134865] 0.03904772401227586
[0.9814877501808776, 0.016806256734246416] 0.017123246552133944
In [10]:
MAPE_mean = []
MAPE_std = []
for i in MAPE_tot:
    print(i, i[1]/i[0])
    MAPE_mean.append(i[0])
    MAPE_std.append(i[1])

[0.19519403842956715, 0.14810321983657285] 0.7587486842740522
[0.19987011487793538, 0.1495318382929695] 0.7481450560244665
[0.06660278937384334, 0.042158541765406204] 0.6329846266463274
[0.0748010166107654, 0.04860910951978934] 0.6498455732591407
[0.04771831837837801, 0.03170268838521743] 0.6643714502643173
[0.051272615779275256, 0.02965868743556692] 0.5784508354956052
[0.04261971466867888, 0.02508246809574713] 0.5885179732134662
[0.0470851870733642, 0.02769338957816019] 0.5881550292029353
[0.17808714305986226, 0.11588936418714697] 0.6507452598540024
[0.17916123818062862, 0.1085848141360262] 0.6060731396963893
[0.0548325290685142, 0.031458211018119596] 0.5737143909377591
[0.06699232661279844, 0.04100126846219039] 0.6120293253758613
[0.04070058669170898, 0.0288365203492443] 0.7085038003916175
[0.051050026889971914, 0.024611010400292437] 0.482095934118439
[0.03819108922130367, 0.030001220359156557] 0.7855555044610077
[0.04631577095675475, 0.023755000853610615] 0.5128922689377398
In [11]:
MAE_mean = []
MAE_std =[]
for i in MAE_tot:
    print(i, i[1]/i[0])
    MAE_mean.append(i[0])
    MAE_std.append(i[1])

[1.024076463758599, 0.7291192462689874] 0.7119773494187631
[1.0409497443151305, 0.6453183761085421] 0.6199323066581998
[0.37429304952389275, 0.25496840105101076] 0.6811999351185789
[0.42368820361761395, 0.28180947789365757] 0.665134113924955
[0.26748321095537647, 0.19213548067068698] 0.71830856218763
[0.3051781248677179, 0.2209770181181784] 0.7240919322574736
[0.25078611939452194, 0.18348659332631734] 0.7316457297130828
[0.28266759962212656, 0.2226890986700329] 0.7878126073441963
[0.9895244069165563, 0.614067229379179] 0.6205680477277621
[0.9689504061684131, 0.5229167962871334] 0.5396734373175394
[0.3111157591399913, 0.20009169567264223] 0.6431422703425573
[0.3695158519876495, 0.2497426703612161] 0.6758645644505756
[0.24136652827431715, 0.2123147644170293] 0.8796363188177027
[0.29232491422445334, 0.17079888149335284] 0.5842775390749274
[0.23671356991761472, 0.2720678065320696] 1.1493544988855495
[0.27056490375016395, 0.18794472072640026] 0.6946382110960921
In [12]:
RMSE_mean = []
RMSE_std = []
for i in RMSE_tot:
    print(i, i[1]/i[0])
    RMSE_mean.append(i[0])
    RMSE_std.append(i[1])

[1.6670870879858384, 0.8905010965821514] 0.5341659131065839
[1.6870298628003126, 0.8323310950036094] 0.49337069447129833
[0.5130898523867832, 0.2921713742306395] 0.5694351055112888
[0.5956324083183224, 0.32841681448849114] 0.5513749921964892
[0.34883546340476174, 0.21981874360059012] 0.6301502188311898
[0.4141666006756527, 0.25998523567703485] 0.6277310513520565
[0.32040170596153306, 0.211825344879328] 0.6611242728675091
[0.379493494471496, 0.27218366077381606] 0.7172287924273231
[1.6801143484754053, 0.812830682415111] 0.4837948578635151
[1.665136502829104, 0.7599101133618877] 0.4563650560003841
[0.46143047852028113, 0.2494602198646781] 0.5406236290776654
[0.5383204655275987, 0.29532293688632355] 0.5486006120850014
[0.3302517363990494, 0.2581530940615623] 0.7816858039154441
[0.40872024771583254, 0.20553341815838705] 0.5028706537222656
[0.31846367971555395, 0.3236163269711548] 1.0161797014347227
[0.36731240855698954, 0.22624789608396179] 0.6159549495558596
In [13]:
df = pd.DataFrame({
    'Files': files,
    'Block':b,
    'Window':windows,
    'Model': m,
    'R2': R2_mean,
    'MAPE': MAPE_mean,
    'RMSE': RMSE_mean,
    'MAE': MAE_mean
})
In [14]:
df_round = df.round(2)
df_round.to_csv("results_mean.csv")
In [15]:
df = pd.DataFrame({
    'Files': files,
    'Block':b,
    'Window':windows,
    'Model': m,
    'R2': R2_std,
    'MAPE': MAPE_std,
    'RMSE': RMSE_std,
    'MAE': MAE_std
})

df_round = df.round(2)
df_round.to_csv("results_std.csv")
In [16]:
R2_cv = []
MAPE_cv = []
MAE_cv = []
RMSE_cv = []
for i in range(0,len(files)):
    R2_cv.append(R2score_tot[i][1]/R2score_tot[i][0])
    MAPE_cv.append(MAPE_tot[i][1]/MAPE_tot[i][0])
    MAE_cv.append(MAE_tot[i][1]/MAE_tot[i][0])
    RMSE_cv.append(RMSE_tot[i][1]/RMSE_tot[i][0])
    print(files[i],R2score_tot[i][1]/R2score_tot[i][0], MAPE_tot[i][1]/MAPE_tot[i][0], MAE_tot[i][1]/MAE_tot[i][0], RMSE_tot[i][1]/RMSE_tot[i][0]) 

predicted_series_rnn_block1_w50.csv 0.1903588908954381 0.7587486842740522 0.7119773494187631 0.5341659131065839
predicted_series_rnn_block1_w20.csv 0.1592875504570695 0.7481450560244665 0.6199323066581998 0.49337069447129833
predicted_series_rnn_block10_w50.csv 0.04133506953373589 0.6329846266463274 0.6811999351185789 0.5694351055112888
predicted_series_rnn_block10_w20.csv 0.029319307421933935 0.6498455732591407 0.665134113924955 0.5513749921964892
predicted_series_rnn_block20_w50.csv 0.025109216687647908 0.6643714502643173 0.71830856218763 0.6301502188311898
predicted_series_rnn_block20_w20.csv 0.02771778888530132 0.5784508354956052 0.7240919322574736 0.6277310513520565
predicted_series_rnn_block30_w50.csv 0.026867875739094497 0.5885179732134662 0.7316457297130828 0.6611242728675091
predicted_series_rnn_block30_w20.csv 0.02896638299161449 0.5881550292029353 0.7878126073441963 0.7172287924273231
predicted_series_lstm_block1_w50.csv 0.16665203276381693 0.6507452598540024 0.6205680477277621 0.4837948578635151
predicted_series_lstm_block1_w20.csv 0.12606880199606035 0.6060731396963893 0.5396734373175394 0.4563650560003841
predicted_series_lstm_block10_w50.csv 0.03319761865708034 0.5737143909377591 0.6431422703425573 0.5406236290776654
predicted_series_lstm_block10_w20.csv 0.04765006568821529 0.6120293253758613 0.6758645644505756 0.5486006120850014
predicted_series_lstm_block20_w50.csv 0.03788713289416336 0.7085038003916175 0.8796363188177027 0.7816858039154441
predicted_series_lstm_block20_w20.csv 0.022644786550697557 0.482095934118439 0.5842775390749274 0.5028706537222656
predicted_series_lstm_block30_w50.csv 0.03904772401227586 0.7855555044610077 1.1493544988855495 1.0161797014347227
predicted_series_lstm_block30_w20.csv 0.017123246552133944 0.5128922689377398 0.6946382110960921 0.6159549495558596
In [17]:
plt.figure(figsize = (30,10))
plt.plot(R2_cv, label = "R2", c = "orange")
plt.plot(MAPE_cv, label = "MAPE", c = "green")
plt.plot(MAE_cv, label = "MAE", c = "red")
plt.plot(RMSE_cv, label = "RMSE", c = "blue")
plt.xlabel("time")
plt.ylabel("distance")
plt.title("Results Graph")
plt.legend()
plt.show()
In [18]:
print('R2:',files[R2_cv.index(min(R2_cv))],min(R2_cv))
print('MAPE:',files[MAPE_cv.index(min(MAPE_cv))],min(MAPE_cv))
print('MAE:',files[MAE_cv.index(min(MAE_cv))],min(MAE_cv))
print('RMSE:',files[RMSE_cv.index(min(RMSE_cv))],min(RMSE_cv))

R2: predicted_series_lstm_block30_w20.csv 0.017123246552133944
MAPE: predicted_series_lstm_block20_w20.csv 0.482095934118439
MAE: predicted_series_lstm_block1_w20.csv 0.5396734373175394
RMSE: predicted_series_lstm_block1_w20.csv 0.4563650560003841
In [19]:
df = pd.DataFrame({
    'Files': files,
    'Block':b,
    'Window':windows,
    'Model': m,
    'R2': R2_cv,
    'MAPE': MAPE_cv,
    'MAE': MAE_cv,
    'RMSE': RMSE_cv
})

df_round = df.round(2)
df_round.to_csv("results_cv.csv")
In [20]:
df['R2_rank'] = df['R2'].rank(ascending=True)
df['MAPE_rank'] = df['MAPE'].rank(ascending=True)
df['MAE_rank'] = df['MAE'].rank(ascending=True)
df['RMSE_rank'] = df['RMSE'].rank(ascending=True)

df['average_rank'] = df[['R2_rank', 'MAPE_rank', 'MAE_rank', 'RMSE_rank']].mean(axis=1)

df_round = df.round(2)
df_sorted = df_round.sort_values('average_rank')
df_sorted[['Block','Window','Model', 'R2', 'MAPE', 'MAE', 'RMSE', 'average_rank']].to_csv("results_ranking.csv")
In [21]:
df_sorted[['Block','Window','Model', 'R2', 'MAPE', 'MAE', 'RMSE', 'average_rank']]
Out[21]:
Block Window Model R2 MAPE MAE RMSE average_rank
13 20 20 lstm 0.02 0.48 0.58 0.50 2.25
9 1 20 lstm 0.13 0.61 0.54 0.46 5.50
10 10 50 lstm 0.03 0.57 0.64 0.54 5.50
15 30 20 lstm 0.02 0.51 0.69 0.62 5.50
3 10 20 rnn 0.03 0.65 0.67 0.55 7.75
5 20 20 rnn 0.03 0.58 0.72 0.63 8.00
8 1 50 lstm 0.17 0.65 0.62 0.48 8.00
1 1 20 rnn 0.16 0.75 0.62 0.49 8.50
11 10 20 lstm 0.05 0.61 0.68 0.55 8.50
6 30 50 rnn 0.03 0.59 0.73 0.66 9.00
2 10 50 rnn 0.04 0.63 0.68 0.57 9.25
4 20 50 rnn 0.03 0.66 0.72 0.63 9.50
7 30 20 rnn 0.03 0.59 0.79 0.72 9.75
0 1 50 rnn 0.19 0.76 0.71 0.53 11.50
12 20 50 lstm 0.04 0.71 0.88 0.78 13.00
14 30 50 lstm 0.04 0.79 1.15 1.02 14.50
In [22]:
df
Out[22]:
Files Block Window Model R2 MAPE MAE RMSE R2_rank MAPE_rank MAE_rank RMSE_rank average_rank
0 predicted_series_rnn_block1_w50.csv 1 50 rnn 0.190359 0.758749 0.711977 0.534166 16.0 15.0 10.0 5.0 11.50
1 predicted_series_rnn_block1_w20.csv 1 20 rnn 0.159288 0.748145 0.619932 0.493371 14.0 14.0 3.0 3.0 8.50
2 predicted_series_rnn_block10_w50.csv 10 50 rnn 0.041335 0.632985 0.681200 0.569435 11.0 9.0 8.0 9.0 9.25
3 predicted_series_rnn_block10_w20.csv 10 20 rnn 0.029319 0.649846 0.665134 0.551375 7.0 10.0 6.0 8.0 7.75
4 predicted_series_rnn_block20_w50.csv 20 50 rnn 0.025109 0.664371 0.718309 0.630150 3.0 12.0 11.0 12.0 9.50
5 predicted_series_rnn_block20_w20.csv 20 20 rnn 0.027718 0.578451 0.724092 0.627731 5.0 4.0 12.0 11.0 8.00
6 predicted_series_rnn_block30_w50.csv 30 50 rnn 0.026868 0.588518 0.731646 0.661124 4.0 6.0 13.0 13.0 9.00
7 predicted_series_rnn_block30_w20.csv 30 20 rnn 0.028966 0.588155 0.787813 0.717229 6.0 5.0 14.0 14.0 9.75
8 predicted_series_lstm_block1_w50.csv 1 50 lstm 0.166652 0.650745 0.620568 0.483795 15.0 11.0 4.0 2.0 8.00
9 predicted_series_lstm_block1_w20.csv 1 20 lstm 0.126069 0.606073 0.539673 0.456365 13.0 7.0 1.0 1.0 5.50
10 predicted_series_lstm_block10_w50.csv 10 50 lstm 0.033198 0.573714 0.643142 0.540624 8.0 3.0 5.0 6.0 5.50
11 predicted_series_lstm_block10_w20.csv 10 20 lstm 0.047650 0.612029 0.675865 0.548601 12.0 8.0 7.0 7.0 8.50
12 predicted_series_lstm_block20_w50.csv 20 50 lstm 0.037887 0.708504 0.879636 0.781686 9.0 13.0 15.0 15.0 13.00
13 predicted_series_lstm_block20_w20.csv 20 20 lstm 0.022645 0.482096 0.584278 0.502871 2.0 1.0 2.0 4.0 2.25
14 predicted_series_lstm_block30_w50.csv 30 50 lstm 0.039048 0.785556 1.149354 1.016180 10.0 16.0 16.0 16.0 14.50
15 predicted_series_lstm_block30_w20.csv 30 20 lstm 0.017123 0.512892 0.694638 0.615955 1.0 2.0 9.0 10.0 5.50
In [23]:
import seaborn as sns
import matplotlib.pyplot as plt
from math import pi
In [24]:
import matplotlib.pyplot as plt
import yfinance as yf

def plot(pred,test,stockname,model,block,w):
    plt.figure(figsize = (30,10))
    plt.plot(pred, label = "prediction", c = "orange")
    plt.plot(test, label = "actual", c = "green")
    plt.xlabel("Time", fontsize=18)
    plt.ylabel("Distance", fontsize=18)
    plt.legend(fontsize=12)
    plt.grid(True)
    stockname = stock[1:-1] 
    img = "results_graphs/{0}_{1}_block{2}_window{3}".format(stockname,model,block,w)
    plt.savefig(img+'.jpeg', dpi=1200, bbox_inches='tight')
    plt.show()
   

def plot_actual(stockname, n,save):
    stockname = stockname[1:-1] 
    s = stockname.split('_')
    print(s)
    s1 = s[0]+'.NS'
    s2 = s[1]+'.NS'
    stock1 = yf.download(s1, start='2003-01-01', end='2023-12-31')['Adj Close']
    stock2 = yf.download(s2, start='2003-01-01', end='2023-12-31')['Adj Close']
    
    # Take the last 3701 prices for each stock
    stock1_1 = stock1[-n:]
    stock2_1 = stock2[-n:]
    
    # Normalize time series individually using standard deviation
    normalized_stock1 = (stock1_1 - stock1_1.mean()) / (stock1_1.std())
    normalized_stock2 = (stock2_1 - stock2_1.mean()) / (stock2_1.std())
    
    # Plot the normalized time series
    plt.figure(figsize=(30, 10))
    plt.plot(normalized_stock1, label=s1)
    plt.plot(normalized_stock2, label=s2)
    plt.xlabel('Date', fontsize=18)
    plt.ylabel('Normalized Price', fontsize=18)
    plt.legend(fontsize=12)
    plt.grid(True)
    if save==1:
        img = "results_graphs/{0}_actual".format(stockname)
        plt.savefig(img+'.jpeg',dpi=1200, bbox_inches='tight')
    plt.show()
In [25]:
stock = '"HDFCBANK_UPL"'
c = 0
for model in models:
    for block in blocks:
        for w in win:
            c+=1
            file = filename.format(model,block,w)
            testfile = "test_data{0}.csv".format(blocks.index(block)+1)
            predicted_series = pd.read_csv(file)
            test_data_file = pd.read_csv(testfile)
            print(file, testfile, w)
            test = test_data_file[stock][int(w):].tolist()
            pred = predicted_series[stock].tolist()
            plot(pred,test, stock, model, block, w)
            if c==1:
                plot_actual(stock, len(test),1)
            else:
                plot_actual(stock, len(test),0)

predicted_series_rnn_block1_w50.csv test_data1.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block1_w20.csv test_data1.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block10_w50.csv test_data2.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block10_w20.csv test_data2.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block20_w50.csv test_data3.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block20_w20.csv test_data3.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block30_w50.csv test_data4.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_rnn_block30_w20.csv test_data4.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block1_w50.csv test_data1.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block1_w20.csv test_data1.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block10_w50.csv test_data2.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block10_w20.csv test_data2.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block20_w50.csv test_data3.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block20_w20.csv test_data3.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block30_w50.csv test_data4.csv 50

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed

predicted_series_lstm_block30_w20.csv test_data4.csv 20

['HDFCBANK', 'UPL']
[*********************100%%**********************]  1 of 1 completed
[*********************100%%**********************]  1 of 1 completed
In [26]:
df = pd.DataFrame({
    'Files': files,
    'R2': R2_cv,
    'MAPE': MAPE_cv,
    'MAE': MAE_cv,
    'RMSE': RMSE_cv
})
df['R2_normalized'] = (df['R2'] - df['R2'].min()) / (df['R2'].max() - df['R2'].min())
df['MAPE_normalized'] = (df['MAPE'].max() - df['MAPE']) / (df['MAPE'].max() - df['MAPE'].min())
df['MAE_normalized'] = (df['MAE'].max() - df['MAE']) / (df['MAE'].max() - df['MAE'].min())
df['RMSE_normalized'] = (df['RMSE'].max() - df['RMSE']) / (df['RMSE'].max() - df['RMSE'].min())
In [27]:
#Classification
In [28]:
epsilons = pd.read_csv("epsilon_test1.csv")
In [29]:
test_data_file = pd.read_csv("test_data3.csv")
print(test_data_file.columns)

Index(['"HDFCBANK_M&M"', '"HDFCBANK_ULTRACEMCO"', '"HDFCBANK_GRASIM"',
       '"HDFCBANK_PIDILITIND"', '"HDFCBANK_LT"', '"HDFCBANK_HINDUNILVR"',
       '"HDFCBANK_ITC"', '"HDFCBANK_RELIANCE"', '"HDFCBANK_ONGC"',
       '"HDFCBANK_UPL"',
       ...
       '"INFY_TATASTEEL"', '"INFY_SUNPHARMA"', '"INFY_CONCOR"',
       '"INFY_BHARTIARTL"', '"TATASTEEL_SUNPHARMA"', '"TATASTEEL_CONCOR"',
       '"TATASTEEL_BHARTIARTL"', '"SUNPHARMA_CONCOR"',
       '"SUNPHARMA_BHARTIARTL"', '"CONCOR_BHARTIARTL"'],
      dtype='object', length=190)
In [30]:
names = test_data_file.columns.tolist()
In [31]:
filename = "predicted_series_lstm_block20_w20.csv"
predicted_series_lstm = pd.read_csv(filename)
In [32]:
#for LSTM
c = 0
for epsilon in epsilons:
    accuracies = []
    precisions_0 = []
    precisions_1 = []
    recalls_0 = []
    recalls_1 = []
    f1_scores_0 = []
    f1_scores_1 = []
    supports_0 = []
    supports_1 = []
    
    for name in names:
        epsilon_list = epsilons[epsilon].tolist()
        threshold = epsilon_list[names.index(name)]
        y_pred = (predicted_series_lstm[name] < threshold).astype(int)
        y_true = (test_data_file[name] < threshold).astype(int)
        y_true = y_true[20:]
        
        # Calculate metrics
        accuracy = accuracy_score(y_true, y_pred)
        precision = precision_score(y_true, y_pred, average=None, zero_division=1)
        recall = recall_score(y_true, y_pred, average=None, zero_division=1)
        f1 = f1_score(y_true, y_pred, average=None, zero_division=1)
        report = classification_report(y_true, y_pred, output_dict=True, zero_division=1)
    
        # Extracting support from classification report
        support_0 = report['0']['support']
        support_1 = report['1']['support'] if '1' in report else 0
    
        # Append results to lists
        accuracies.append(accuracy)
        precisions_0.append(precision[0])
        recalls_0.append(recall[0])
        f1_scores_0.append(f1[0])
        supports_0.append(support_0)
        if support_1 > 0:
            precisions_1.append(precision[1])
            recalls_1.append(recall[1])
            f1_scores_1.append(f1[1])
            supports_1.append(support_1)
        else:
            precisions_1.append(None)
            recalls_1.append(None)
            f1_scores_1.append(None)
            supports_1.append(0)

    per_class1 = []
    for i in range(len(supports_1)):
        per1 = (supports_1[i]/(supports_1[i] + supports_0[i]))
        per_class1.append(per1)

    df_results = pd.DataFrame({
        "stock_pair": names,
        "accuracy": accuracies,
        "precision": precisions_1,
        "recall": recalls_1,
        "f1": f1_scores_1,
        "%class 1": per_class1
    })
    
    # Save dataframe to CSV
    df_results.to_csv('lstm_block20_w20_classification_results_{0}.csv'.format(epsilon), index=False)
In [33]:
from prettytable import PrettyTable
dfs = []

for epsilon in epsilons:
    df = pd.read_csv(f'lstm_block20_w20_classification_results_{epsilon}.csv')
    df['epsilon'] = epsilon
    dfs.append(df)

df_results = pd.concat(dfs, ignore_index=True)
In [34]:
summary = df_results.groupby('epsilon').agg(['mean', 'std', 'min', 'max', 'median'])
summary.columns = ['_'.join(col).strip() for col in summary.columns.values]

C:\Users\sanja\AppData\Local\Temp\ipykernel_48892\3977975528.py:1: FutureWarning: ['stock_pair'] did not aggregate successfully. If any error is raised this will raise in a future version of pandas. Drop these columns/ops to avoid this warning.
  summary = df_results.groupby('epsilon').agg(['mean', 'std', 'min', 'max', 'median'])
In [35]:
def generate_summary_table(df, epsilon):
    # Filter for the specific epsilon value
    epsilon_df = df[df['epsilon'] == epsilon]
    # Select only numeric columns for aggregation
    numeric_cols = epsilon_df.select_dtypes(include=['number']).columns
    # Calculate summary statistics
    summary = epsilon_df[numeric_cols].agg(['mean', 'std', 'min', 'max', 'median']).T
    summary.columns = ['Mean', 'Std Dev', 'Min', 'Max', 'Median']
    
    return summary
In [36]:
pd.set_option('display.float_format', '{:.2f}'.format)

for epsilon in epsilons:
    summary_table = generate_summary_table(df_results, epsilon)
    print(f"\nSummary Statistics for Epsilon = {epsilon}")
    display(summary_table)
    summary_table_rounded = summary_table.round(2)
    print(summary_table['Mean']['accuracy'])
    summary_table_rounded.to_csv('lstm_block20_w20_summary_results_{0}.csv'.format(epsilon), index=True)

Summary Statistics for Epsilon = RR 1% (Testing)
Mean Std Dev Min Max Median
accuracy 1.00 0.00 0.97 1.00 1.00
precision 0.91 0.16 0.57 1.00 1.00
recall 0.45 0.41 0.00 1.00 0.54
f1 0.48 0.41 0.00 0.90 0.63
%class 1 0.00 0.00 0.00 0.03 0.00 
0.9996822765575356

Summary Statistics for Epsilon = RR 5%(Testing)
Mean Std Dev Min Max Median
accuracy 1.00 0.01 0.94 1.00 1.00
precision 0.81 0.31 0.00 1.00 0.97
recall 0.40 0.36 0.00 1.00 0.37
f1 0.55 0.36 0.00 1.00 0.62
%class 1 0.01 0.01 0.00 0.09 0.00 
0.9971025003453515

Summary Statistics for Epsilon = RR 10%(Testing)
Mean Std Dev Min Max Median
accuracy 0.99 0.01 0.96 1.00 1.00
precision 0.92 0.13 0.25 1.00 0.99
recall 0.52 0.37 0.00 1.00 0.65
f1 0.56 0.37 0.00 1.00 0.74
%class 1 0.02 0.03 0.00 0.15 0.00 
0.9942602569415665

Summary Statistics for Epsilon = RR 20%(Testing)
Mean Std Dev Min Max Median
accuracy 0.99 0.02 0.86 1.00 0.99
precision 0.91 0.16 0.00 1.00 0.97
recall 0.70 0.32 0.00 1.00 0.84
f1 0.74 0.29 0.00 1.00 0.88
%class 1 0.05 0.06 0.00 0.27 0.02 
0.9882200580190635

Summary Statistics for Epsilon = RR 25%_Testing
Mean Std Dev Min Max Median
accuracy 0.98 0.02 0.86 1.00 0.99
precision 0.94 0.09 0.50 1.00 0.97
recall 0.71 0.31 0.00 1.00 0.85
f1 0.75 0.29 0.00 1.00 0.89
%class 1 0.07 0.08 0.00 0.33 0.05 
0.9845351567896118

Summary Statistics for Epsilon = RR 30%_Testing
Mean Std Dev Min Max Median
accuracy 0.98 0.02 0.88 1.00 0.99
precision 0.94 0.10 0.19 1.00 0.97
recall 0.79 0.24 0.00 1.00 0.90
f1 0.83 0.20 0.00 1.00 0.91
%class 1 0.10 0.10 0.00 0.41 0.06 
0.982836027075563

Summary Statistics for Epsilon = RR 35%_Testing
Mean Std Dev Min Max Median
accuracy 0.98 0.02 0.88 1.00 0.99
precision 0.94 0.12 0.00 1.00 0.97
recall 0.80 0.27 0.00 1.00 0.93
f1 0.83 0.24 0.00 1.00 0.93
%class 1 0.13 0.12 0.00 0.51 0.10 
0.9808606161071973

Summary Statistics for Epsilon = RR 40%_Testing
Mean Std Dev Min Max Median
accuracy 0.98 0.02 0.90 1.00 0.99
precision 0.94 0.12 0.00 1.00 0.98
recall 0.83 0.25 0.00 1.00 0.94
f1 0.86 0.22 0.00 1.00 0.94
%class 1 0.16 0.14 0.00 0.57 0.13 
0.9811576184555879

Summary Statistics for Epsilon = RR 45%_Testing
Mean Std Dev Min Max Median
accuracy 0.98 0.02 0.90 1.00 0.98
precision 0.95 0.09 0.25 1.00 0.97
recall 0.86 0.22 0.00 1.00 0.94
f1 0.88 0.20 0.00 1.00 0.95
%class 1 0.19 0.16 0.00 0.63 0.17 
0.9790406133443845
In [37]:
def calculate_cv(summary_table):
    summary_table['CV'] = summary_table['Std Dev'] / summary_table['Mean']
    return summary_table

pd.set_option('display.float_format', '{:.4f}'.format)

# Initialize an empty DataFrame to store CV values for each epsilon
cv_df = pd.DataFrame()

for epsilon in epsilons:
    summary_table = generate_summary_table(df_results, epsilon)
    summary_table = calculate_cv(summary_table)
    print(f"\nSummary Statistics for Epsilon = {epsilon}")
    display(summary_table)
    
    # Extract the CV values and store them in a DataFrame
    cv_values = summary_table['CV'].to_frame().T
    cv_values['epsilon'] = epsilon
    cv_df = pd.concat([cv_df, cv_values], ignore_index=True)

# Set 'epsilon' as the index
cv_df.set_index('epsilon', inplace=True)

# Drop the first 3 rows
cv_df = cv_df.iloc[3:]

# Display the CV DataFrame
print("\nCoefficient of Variation (CV) for Each Epsilon")
display(cv_df)

Summary Statistics for Epsilon = RR 1% (Testing)
Mean Std Dev Min Max Median CV
accuracy 0.9997 0.0025 0.9665 1.0000 1.0000 0.0025
precision 0.9062 0.1570 0.5714 1.0000 1.0000 0.1733
recall 0.4550 0.4058 0.0000 1.0000 0.5357 0.8919
f1 0.4817 0.4144 0.0000 0.9000 0.6303 0.8602
%class 1 0.0004 0.0027 0.0000 0.0335 0.0000 6.8160 
Summary Statistics for Epsilon = RR 5%(Testing)
Mean Std Dev Min Max Median CV
accuracy 0.9971 0.0077 0.9403 1.0000 1.0000 0.0077
precision 0.8088 0.3101 0.0000 1.0000 0.9713 0.3833
recall 0.4042 0.3621 0.0000 1.0000 0.3670 0.8959
f1 0.5498 0.3600 0.0000 1.0000 0.6216 0.6548
%class 1 0.0056 0.0149 0.0000 0.0879 0.0000 2.6749 
Summary Statistics for Epsilon = RR 10%(Testing)
Mean Std Dev Min Max Median CV
accuracy 0.9943 0.0086 0.9560 1.0000 1.0000 0.0086
precision 0.9243 0.1312 0.2500 1.0000 0.9917 0.1420
recall 0.5232 0.3726 0.0000 1.0000 0.6548 0.7122
f1 0.5596 0.3678 0.0000 1.0000 0.7429 0.6574
%class 1 0.0170 0.0305 0.0000 0.1483 0.0000 1.7981 
Summary Statistics for Epsilon = RR 20%(Testing)
Mean Std Dev Min Max Median CV
accuracy 0.9882 0.0168 0.8648 1.0000 0.9938 0.0170
precision 0.9129 0.1626 0.0000 1.0000 0.9674 0.1781
recall 0.6976 0.3230 0.0000 1.0000 0.8443 0.4631
f1 0.7446 0.2916 0.0000 1.0000 0.8832 0.3916
%class 1 0.0503 0.0648 0.0000 0.2677 0.0233 1.2894 
Summary Statistics for Epsilon = RR 25%_Testing
Mean Std Dev Min Max Median CV
accuracy 0.9845 0.0206 0.8615 1.0000 0.9915 0.0210
precision 0.9411 0.0897 0.5000 1.0000 0.9684 0.0954
recall 0.7135 0.3134 0.0000 1.0000 0.8542 0.4392
f1 0.7519 0.2902 0.0000 0.9951 0.8872 0.3860
%class 1 0.0721 0.0829 0.0000 0.3307 0.0456 1.1495 
Summary Statistics for Epsilon = RR 30%_Testing
Mean Std Dev Min Max Median CV
accuracy 0.9828 0.0208 0.8753 1.0000 0.9895 0.0211
precision 0.9400 0.0974 0.1875 1.0000 0.9686 0.1036
recall 0.7943 0.2385 0.0000 1.0000 0.8994 0.3003
f1 0.8310 0.2007 0.0000 0.9951 0.9139 0.2416
%class 1 0.0979 0.1029 0.0000 0.4147 0.0650 1.0507 
Summary Statistics for Epsilon = RR 35%_Testing
Mean Std Dev Min Max Median CV
accuracy 0.9809 0.0206 0.8825 1.0000 0.9875 0.0210
precision 0.9421 0.1198 0.0000 1.0000 0.9714 0.1272
recall 0.7982 0.2688 0.0000 1.0000 0.9281 0.3367
f1 0.8318 0.2355 0.0000 1.0000 0.9282 0.2831
%class 1 0.1269 0.1220 0.0000 0.5138 0.1004 0.9617 
Summary Statistics for Epsilon = RR 40%_Testing
Mean Std Dev Min Max Median CV
accuracy 0.9812 0.0193 0.8990 1.0000 0.9869 0.0196
precision 0.9436 0.1245 0.0000 1.0000 0.9752 0.1320
recall 0.8304 0.2544 0.0000 1.0000 0.9372 0.3063
f1 0.8648 0.2169 0.0000 1.0000 0.9434 0.2508
%class 1 0.1564 0.1405 0.0000 0.5728 0.1306 0.8985 
Summary Statistics for Epsilon = RR 45%_Testing
Mean Std Dev Min Max Median CV
accuracy 0.9790 0.0194 0.8963 1.0000 0.9833 0.0198
precision 0.9488 0.0868 0.2500 1.0000 0.9747 0.0915
recall 0.8572 0.2238 0.0000 1.0000 0.9431 0.2611
f1 0.8762 0.1989 0.0000 0.9976 0.9451 0.2270
%class 1 0.1864 0.1585 0.0000 0.6293 0.1703 0.8502 
Coefficient of Variation (CV) for Each Epsilon
accuracy precision recall f1 %class 1
epsilon
RR 20%(Testing) 0.0170 0.1781 0.4631 0.3916 1.2894
RR 25%_Testing 0.0210 0.0954 0.4392 0.3860 1.1495
RR 30%_Testing 0.0211 0.1036 0.3003 0.2416 1.0507
RR 35%_Testing 0.0210 0.1272 0.3367 0.2831 0.9617
RR 40%_Testing 0.0196 0.1320 0.3063 0.2508 0.8985
RR 45%_Testing 0.0198 0.0915 0.2611 0.2270 0.8502
In [38]:
rank_df = cv_df.rank(ascending=True)

# Calculate average rank for each epsilon
cv_df['average_rank'] = rank_df.mean(axis=1)

# Sort epsilons by average rank
ranked_epsilons = cv_df.sort_values(by='average_rank')

# Display the ranked epsilons
print("Ranked Epsilons Based on Average Rank")
display(ranked_epsilons)
ranked_epsilons_rounded = ranked_epsilons.round(2)
ranked_epsilons_rounded.to_csv('ranked_epsilons.csv')

Ranked Epsilons Based on Average Rank
accuracy precision recall f1 %class 1 average_rank
epsilon
RR 45%_Testing 0.0198 0.0915 0.2611 0.2270 0.8502 1.4000
RR 40%_Testing 0.0196 0.1320 0.3063 0.2508 0.8985 3.0000
RR 30%_Testing 0.0211 0.1036 0.3003 0.2416 1.0507 3.4000
RR 35%_Testing 0.0210 0.1272 0.3367 0.2831 0.9617 4.0000
RR 25%_Testing 0.0210 0.0954 0.4392 0.3860 1.1495 4.2000
RR 20%(Testing) 0.0170 0.1781 0.4631 0.3916 1.2894 5.0000
In [39]:
df2 = df_results[df_results['epsilon'] == 'RR 45%_Testing']
df2
Out[39]:
stock_pair accuracy precision recall f1 %class 1 epsilon
1520 "HDFCBANK_M&M" 0.9882 0.9203 0.9478 0.9338 0.0879 RR 45%_Testing
1521 "HDFCBANK_ULTRACEMCO" 0.9731 0.9178 0.9877 0.9515 0.2671 RR 45%_Testing
1522 "HDFCBANK_GRASIM" 0.9941 0.9986 0.9889 0.9937 0.4711 RR 45%_Testing
1523 "HDFCBANK_PIDILITIND" 0.9711 0.9632 0.7706 0.8562 0.1115 RR 45%_Testing
1524 "HDFCBANK_LT" 0.9928 0.9924 0.9286 0.9594 0.0919 RR 45%_Testing
... ... ... ... ... ... ... ...
1705 "TATASTEEL_CONCOR" 0.9770 0.9655 0.8924 0.9275 0.1647 RR 45%_Testing
1706 "TATASTEEL_BHARTIARTL" 0.9731 0.8953 0.9636 0.9282 0.1804 RR 45%_Testing
1707 "SUNPHARMA_CONCOR" 0.9921 0.8913 0.8542 0.8723 0.0315 RR 45%_Testing
1708 "SUNPHARMA_BHARTIARTL" 0.9777 0.9289 0.9015 0.9150 0.1332 RR 45%_Testing
1709 "CONCOR_BHARTIARTL" 0.9856 0.9778 0.6769 0.8000 0.0427 RR 45%_Testing

190 rows × 7 columns

In [40]:
len(df2[df2['f1_1'] == 0.0])

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\pandas\core\indexes\base.py:3802, in Index.get_loc(self, key, method, tolerance)
   3801 try:
-> 3802     return self._engine.get_loc(casted_key)
   3803 except KeyError as err:

File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\pandas\_libs\index.pyx:138, in pandas._libs.index.IndexEngine.get_loc()

File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\pandas\_libs\index.pyx:165, in pandas._libs.index.IndexEngine.get_loc()

File pandas\_libs\hashtable_class_helper.pxi:5745, in pandas._libs.hashtable.PyObjectHashTable.get_item()

File pandas\_libs\hashtable_class_helper.pxi:5753, in pandas._libs.hashtable.PyObjectHashTable.get_item()

KeyError: 'f1_1'

The above exception was the direct cause of the following exception:

KeyError                                  Traceback (most recent call last)
Cell In[40], line 1
----> 1 len(df2[df2['f1_1'] == 0.0])

File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\pandas\core\frame.py:3807, in DataFrame.__getitem__(self, key)
   3805 if self.columns.nlevels > 1:
   3806     return self._getitem_multilevel(key)
-> 3807 indexer = self.columns.get_loc(key)
   3808 if is_integer(indexer):
   3809     indexer = [indexer]

File ~\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\pandas\core\indexes\base.py:3804, in Index.get_loc(self, key, method, tolerance)
   3802     return self._engine.get_loc(casted_key)
   3803 except KeyError as err:
-> 3804     raise KeyError(key) from err
   3805 except TypeError:
   3806     # If we have a listlike key, _check_indexing_error will raise
   3807     #  InvalidIndexError. Otherwise we fall through and re-raise
   3808     #  the TypeError.
   3809     self._check_indexing_error(key)

KeyError: 'f1_1'
In [ ]:
df2[df2['stock_pair'] == '"HDFCBANK_UPL"']
In [ ]:
df2[df2['f1_1'] <= 0.4]
In [ ]:
len(df2[df2['f1_1'] <= 0.4])
In [ ]:
filename = "predicted_series_lstm_block20_w20.csv"
predicted_series_lstm = pd.read_csv(filename)

test_data_file = pd.read_csv('test_data3.csv')
In [ ]:
def plot2(pred,test,stockname,block,w):
    plt.figure(figsize = (30,10))
    plt.plot(pred, label = "prediction", c = "orange")
    plt.plot(test, label = "actual", c = "green")
    plt.xlabel("Time", fontsize=18)
    plt.ylabel("Distance", fontsize=18)
    plt.legend(fontsize=14)
    plt.xticks(fontsize=12)
    plt.yticks(fontsize=12)
    plt.axvspan(215, 375, color='grey', alpha=0.2)
    plt.axvspan(125, 215, color='red', alpha=0.2)
    plt.axvspan(50, 125, color='green', alpha=0.2)
    plt.grid(True)
    plt.savefig('ComparisonGraph_Prediction.jpeg', dpi=1200, bbox_inches='tight')
    plt.show()
   

def plot_actual2(stockname, n):
    stockname = stockname[1:-1] 
    s = stockname.split('_')
    print(s)
    s1 = s[0]+'.NS'
    s2 = s[1]+'.NS'
    stock1 = yf.download(s1, start='2003-01-01', end='2023-12-31')['Adj Close']
    stock2 = yf.download(s2, start='2003-01-01', end='2023-12-31')['Adj Close']
    
    # Take the last 3701 prices for each stock
    stock1_1 = stock1[-n:]
    stock2_1 = stock2[-n:]
    
    # Normalize time series individually using standard deviation
    normalized_stock1 = (stock1_1 - stock1_1.mean()) / (stock1_1.std())
    normalized_stock2 = (stock2_1 - stock2_1.mean()) / (stock2_1.std())
    
    # Plot the normalized time series
    plt.figure(figsize=(30, 10))
    plt.plot(normalized_stock1, label=s[0])
    plt.plot(normalized_stock2, label=s[1])
    plt.xlabel('Date', fontsize=18)
    plt.ylabel('Normalized Price', fontsize=18)
    plt.legend(fontsize=14)
    plt.xticks(fontsize=12)
    plt.yticks(fontsize=12)
    plt.axvspan('2018-11-01', '2019-08-01', color='grey', alpha=0.2)
    plt.axvspan('2018-06-01', '2018-11-01', color='red', alpha=0.2)
    plt.axvspan('2018-02-01', '2018-06-01', color='green', alpha=0.2)
    plt.grid(True)
    plt.savefig('ComparisonGraph_Actual.jpeg', dpi=1200, bbox_inches='tight')
    plt.show()
In [ ]:
stock = '"HDFCBANK_UPL"'
test = test_data_file[stock][int(20):].tolist()
pred = predicted_series_lstm[stock].tolist()
plot2(pred,test, stock, 20, 20)
plot_actual2(stock, len(test))
In [ ]:
import matplotlib.pyplot as plt
import yfinance as yf

def plot2(ax, pred, test):
    ax.plot(pred, label="prediction", c="orange")
    ax.plot(test, label="actual", c="green")
    ax.set_xlabel("Time", fontsize=18)
    ax.set_ylabel("Distance", fontsize=18)
    ax.legend(fontsize=14)
    ax.tick_params(axis='x', labelsize=12)
    ax.tick_params(axis='y', labelsize=12)
    ax.axvspan(215, 375, color='grey', alpha=0.2)
    ax.axvspan(125, 215, color='red', alpha=0.2)
    ax.axvspan(50, 125, color='green', alpha=0.2)
    ax.grid(True)
    ax.text(0.5, -0.15, '(a)', transform=ax.transAxes, fontsize=20, va='top', ha='center')

def plot_actual2(ax, stockname, n):
    stockname = stockname[1:-1]
    s = stockname.split('_')
    s1 = s[0]+'.NS'
    s2 = s[1]+'.NS'
    stock1 = yf.download(s1, start='2003-01-01', end='2023-12-31')['Adj Close']
    stock2 = yf.download(s2, start='2003-01-01', end='2023-12-31')['Adj Close']
    
    stock1_1 = stock1[-n:]
    stock2_1 = stock2[-n:]
    
    normalized_stock1 = (stock1_1 - stock1_1.mean()) / stock1_1.std()
    normalized_stock2 = (stock2_1 - stock2_1.mean()) / stock2_1.std()
    
    ax.plot(normalized_stock1, label=s[0])
    ax.plot(normalized_stock2, label=s[1])
    ax.set_xlabel('Date', fontsize=18)
    ax.set_ylabel('Normalized Price', fontsize=18)
    ax.legend(fontsize=14)
    ax.tick_params(axis='x', labelsize=12)
    ax.tick_params(axis='y', labelsize=12)
    ax.axvspan('2018-11-01', '2019-08-01', color='grey', alpha=0.2)
    ax.axvspan('2018-06-01', '2018-11-01', color='red', alpha=0.2)
    ax.axvspan('2018-02-01', '2018-06-01', color='green', alpha=0.2)
    ax.grid(True)
    ax.text(0.5, -0.15, '(b)', transform=ax.transAxes, fontsize=20, va='top', ha='center')

# Create subplots
fig, axes = plt.subplots(2, 1, figsize=(30, 20))

# Plot prediction vs actual
plot2(axes[0], pred, test)

# Plot actual stock prices
plot_actual2(axes[1], stock, len(test))

# Adjust layout to add spacing between plots
plt.tight_layout(pad=6.0)

# Save and show the figure
plt.savefig('ComparisonGraph_Combined.jpeg', dpi=1200, bbox_inches='tight')
plt.show()
In [ ]:
In [ ]:
In [ ]: